ScholarLens Back to search
2002•Unpublished venueRequires access

MIMD versus SIMD computation: experience with non-numeric parallel algorithms

Clay P. Breshears, Michael Allen Langston

Open publisher page 1 citations

Abstract

The authors study the design and implementation issues for nonnumeric parallel algorithms on multiple instruction/multiple data (MIMD) and single instruction, multiple data (SIMD) machines. Prototypical examples considered are time-space optimal merging and sorting routines. The representative MIMD and SIMD machines are the Sequence Symmetry SSI and the Connection Machine CM-2. The conversion of shared-memory parallel codes to the data parallel paradigm is described. The experiences are highlighted with examples obtained during the process of MIMD to SIMD program transformation. Unexpected events can occur when algorithms designed with one architectural style in mind are modified for execution on another. Timings and other measurements are useful in identifying important bottlenecks. Some relative strengths and weaknesses of the two competing models that have become evident during this transformation process are discussed.>

About this research paper

What this paper is about

The authors study the design and implementation issues for nonnumeric parallel algorithms on multiple instruction/multiple data (MIMD) and single instruction, multiple data (SIMD) machines. Prototypical examples considered are time-space optimal merging and sorting routines. The representative MIMD and SIMD machines are the Sequence Symmetry SSI and the Connection Machine CM-2. The conversion of shared-memory parallel codes to the data parallel paradigm is described. The experiences are highlighted with examples obtained during the process of MIMD to SIMD program transformation. Unexpected events can occur when algorithms designed with one architectural style in mind are modified for execution on another. Timings and other measurements are useful in identifying important bottlenecks. Some relative strengths and weaknesses of the two competing models that have become evident during this transformation process are discussed.>

Why it matters

OpenAlex reports 1 citations for this work. Citation counts describe recorded attention and do not establish research quality.

Key contribution

A contribution statement is not available in the OpenAlex record.

Method / approach

Method details are not available in the OpenAlex metadata.

Main findings

Findings are not separately available in the OpenAlex metadata.

Limitations

Limitations are not available in the OpenAlex metadata.

Applications

Application details are not available in the OpenAlex metadata.

Available abstract

The authors study the design and implementation issues for nonnumeric parallel algorithms on multiple instruction/multiple data (MIMD) and single instruction, multiple data (SIMD) machines. Prototypical examples considered are time-space optimal merging and sorting routines. The representative MIMD and SIMD machines are the Sequence Symmetry SSI and the Connection Machine CM-2. The conversion of shared-memory parallel codes to the data parallel paradigm is described. The experiences are highlighted with examples obtained during the process of MIMD to SIMD program transformation. Unexpected events can occur when algorithms designed with one architectural style in mind are modified for execution on another. Timings and other measurements are useful in identifying important bottlenecks. Some relative strengths and weaknesses of the two competing models that have become evident during this transformation process are discussed.>

Key concepts: MIMD, SIMD, Computer science, Parallel computing, Transformation (genetics), Parallel algorithm, Computation, Process (computing)

Related papers

Related paper 1Related paper 2
Back to paper searchBrowse research topicsOriginal source
MIMD versus SIMD computation: experience with non-numeric parallel algorithms — Research Paper | ScholarLens